Client/server speech recognition systems where a client computer sends speech data to one or more server side speech recognition servers for recognition and receives a speech recognition result corresponding to the speech data in return have become more common in recent years. Such systems allow speech recognition operations to be performed on speech data from computers that need not have capabilities locally to perform automatic speech recognition.